Measuring The Suburbanization of world cities

with Remote Sensing Data

Christopher Chase-Dunn

Institute for Research on World-Systems

University of California-Riverside

John R. Weeks


San Diego State University

 v. 1-25-04, 11172 words

ASTER image of Los Angeles.



          The problem of sustainable urbanization is crucial for the human encounter with the consequences of our ballooning environmental footprint. Over half of the human population of the Earth now lives in very large cities, and these have spread rapidly over the land as population densities within cities have decreased and cities have spread into huge city-regions. Our research project is developing a methodology for measuring the rates and the nature of the areal expansion of world cities and the patterns of decreasing population density in order to know whether or not urban sprawl is accelerating or slowing down.  We are also studying changes in the world city-size distribution over the past three decades.  The size distribution of world cities has been flattening since the 1950s as megacities in the non-core countries have caught up in terms of overall population size with the global cities of the core.  We are examining this trend closely to see if it has leveled off or accelerated.  And we are studying differences among the cities of the core and the non-core with respect to the rates and nature of urban sprawl and the changing structure of the built environment.


            Our project is developing measures for using satellite data to study changes in world cities over the last three decades. Global coverage, large spatial scale, easily standardized spatial format, and relatively low cost make satellite data attractive for studying world cities and city-regions. The main problem is that the satellite technology has changed very rapidly and so the comparison of early with recent satellite data requires a sophisticated approach to measurement error. In order to develop algorithms for solving this problem our project is focusing on five cities in North America, Africa, and the Middle East where we already have done research and have access to ancillary data on land-use patterns that will enable us to develop techniques for decomposing change into that which is due to measurement error and that which is real change as indicated by the classification of data from the satellite imagery.


The nature of the human built environment has huge consequences for humanity itself and increasingly large impacts on the global biosphere. Social scientists have long studied the emergence of sedentism and the growth of settlements, but only recently has attention been drawn to the relationship between cities and the environment. Our study of the global city system and the suburbanization of world cities has important implications for theories of urban growth and the development of settlement systems, and for the relationship between urbanization and the environment.


The project is producing an educational World Cities Time-Map, a web-accessible spatio-temporal GIS that presents animations of growth and suburbanization of the largest cities on Earth over the period from 1950 to 2005. The World Cities Time-Map is designed to provide information that will be valuable for public and private decision-makers on the issues of sprawl and sustainable urbanization.


Measuring the Suburbanization of World Cities with Remote Sensing Data


            Changes in the global city size distribution, especially its flattening as megacities have emerged in the non-core, have important implications for theories of urban growth, globalization and the future of global inequalities. Whether or not rates of growth and suburbanization are accelerating, steady, or decreasing, and how these may vary between core, peripheral and semiperipheral societies in the world-system, will have important implications for theories of urban growth and for the future of urban sustainability.  However, a fundamental problem faced by all researchers is the lack of consistent data on the spatial extent of cities and, related to that, on the changing population size and density of urban places in different parts of the globe. This research project is developing new methods for using remotely sensed data from satellites to study the growth of world cities and city-regions, patterns of urban multinucleation, rates of suburbanization and trends in the global size distribution of cities.  The first stage of our research is designed to test alternative strategies of using remotely sensed data to capture information about city growth between 1980 and 2005 that can then be used to test theories of urban growth, city systems, and urban sustainability.  We propose to do this for five cities—two core and three non-core cities, building on work already underway by the researchers.  The second stage is to extend the analysis beyond the five cities, ultimately producing data for the 251 world cities that we have identified.  The third stage is to develop techniques by which urban spatial extent and urban population dynamics can be modeled backwards in time, allowing us to “reverse engineer” the theories of city growth and city system development.


            The objectives of our research are the following: (1) define the characteristics of urban places and the population living in urban places that can be measured by the classification of data derived from remotely sensed (RS) imagery; (2) test alternative methods of classifying RS data to most accurately capture the changes taking place in the spatial and demographic nature of cities; (3) test alternative methods of evaluating the measurement error inherent in using different remote sensors to measure changes over time and differences among places; (4) use the results of (2) and (3) to develop an algorithm to decompose observed change in an urban place as derived from imagery into its three constituent components of classification error, remote sensor error, and real change; (5) use the measured real change to test theories about city growth, especially suburbanization, and about its historical and potential future effect on world city systems and urban sustainability; (6) disseminate the results of both the methodological and theory-testing components of the research to a wide range of potential users of this information, through traditional academic outlets, as well through a website that incorporates time-mapping in a way that will make the results accessible to a wider lay audience, including high school students; and (7) use the results to lay out the plan to move from stage one to stage two of our overall research program.



            Systems of cities represent human interaction networks and their connections with the built and natural environments. Logically, the study of city systems is a subcategory of the more general topic of settlement systems. Once humans began living in fairly permanent hamlets and villages it became possible to study the interactions of these settlements with one another. Settlements are rarely ever intelligible without knowing their relations with the rural and nomadic populations that interact with them. Archaeologists, ethnographers and, of course, geographers, map the ways in which human habitations are spread across space, and this is a fundamental window on the lives of the people in all social systems.


The spatial aspect of population density is one of the most fundamental variables for understanding the constraints and possibilities of human social organization. The “settlement size distribution” – the relative population sizes of the settlements within a region-- is an important and easily ascertained aspect of all sedentary social systems. And the functional differences among settlements are a basic feature of the division of labor that links households and communities with larger polities and interpolity systems. The emergence of social hierarchies is often related to size hierarchies of settlements. And the building of monumental architecture in large settlements has been closely associated with the emergence of more hierarchical social structures – complex chiefdoms and early states.

      The role of city systems in the reproduction and transformation of human social institutions has been altered by the emergence and predominance of capitalist accumulation, and by the control over mortality that has emerged within the same historical-economic context. By their very nature urban populations require a “surplus” of agricultural production in order to survive because urban places are inherently non-agricultural.  Furthermore, prior to the control over mortality, the crowding of people into urban environments almost invariably increased the spread of disease and raised the level of mortality compared to rural places, and so the rate of natural increase in most urban places for most of human history was negative. For this reason, urban places needed constantly to recruit from the countryside in order to sustain population size. Thus, whereas most of the important cities of agrarian tributary states were centers of control and coordination for the extraction of resources and labor from vast empires by means of institutionalized coercion, the most important cities in the modern world have increasingly supplemented the coordination of force with the manipulations of money and the production of commodities. Obviously military force is still an important element of power in the modern world-system, but the uses of military power have been fundamentally altered by the predominance of capitalist accumulation.  Furthermore, recruitment of labor has been fundamentally turned on its head by the dramatic decline in mortality taking place over the past centuries, which has (a) lowered death rates in cities more than in rural areas, producing positive rates of natural increase, and (b) lowered death rates in rural areas to levels well below fertility, producing high rates of rural population growth, leading to a redundancy of the rural population which then produces a steady flow of migrants to the cities in search of jobs (Davis 1972; 1973; Weeks 2002).

      The long rise of capitalism was promoted by semiperipheral capitalist city-states, usually maritime coordinators of trade protected by naval power. The Italian city-states of Venice and Genoa are perhaps the most famous of these, but the Phoenician city-states of the Mediterranean exploited a similar interstitial niche within a larger system dominated by tributary empires. The niche pioneered by capitalist city-states expanded and became more predominant in the guise of core capitalist nation-states in a series of transformations from Venice and Genoa to the Dutch Republic (led by Amsterdam) and eventually the Pax Britannica coordinated by the great world city of the nineteenth century, London (Chase-Dunn and Willard 1994). Within London the functions mentioned above were spatially separated: empire in Westminster and money in the City. In the twentieth century hegemony of the United States these global functions became located in separate cities (Washington, DC and New York). 

        Thus the role of cities in world-systems changed greatly as capitalism became the predominant mode of accumulation over the last 500 years, and as death rates dropped precipitously over the past 200 years.  In earlier world-systems the biggest cities were empire-cities based on the ability of states to extract resources using institutionalized coercion (armies, bureaucracies, etc.) Capitalist cities existed, but they were in the semiperipheral spaces between the large tributary empires. With the rise of Europe we have capitalist cities becoming the most important cities in the whole world-system. This is especially obvious with the rise of Amsterdam, London and New York – the world cities of the capitalist era.



        The great wave of globalization in the second half of the twentieth century has been heralded (and protested) by the public as well as by social scientists as a new stage of global capitalism with allegedly unique qualities based on new technologies of communication and information processing. Some students of globalization claim that they do not need to know anything about what happened before 1960 because so much has changed that the past is entirely non-comparable with the present. Most of the burgeoning literature on global cities and the world city system shares this breathless presentism. All social systems have exhibited waves of spatial expansion and intensification of large interaction networks followed by contractions. The real question is which aspects of the most current wave are unique and which are functional repetitions of earlier pulsations. 

        Clearly one of the unique elements of globalization is that it too is closely related to the drop in mortality, because that is what set in motion the enormous increase in worldwide population growth.  Control over mortality, and subsequently control over fertility as well, took place not coincidentally within the realm of the capitalist world cities.  Those cities created a social structure within which scientific knowledge could flourish by stimulating a strong market for the products of scientific research.  These included discoveries about the causes of communicable disease, followed by discoveries of how to control communicable disease.  Initially the benefits of these improvements in life expectancy were limited to people living within the core countries, and it was in the countries of Europe and North America that population first began to grow rapidly in the modern world.  Between World Wars I and II these technologies began to be transferred beyond the core, especially from the U.S. to Latin America. However, after World War II, death control technology was spread globally, especially through the work of various United Nations agencies, but funded by the governments of core countries.  Since declines in mortality initially affect infants more than any other age group, there tends to be somewhat delayed reaction in the realization of the effects of a mortality decline until those people who would otherwise have died reach an age where they must be educated, clothed, fed, and jobs and homes created for them on a scale never before imagined. The response of core countries to this enormous increase in demand in the rest of the world has been a large part of what is seen as globalization.


World Cities 

According to the theorists of global capitalism it was during the 1960’s that the organization of economic activity entered a new period expressed by the altered structure of the world economy: the dismantling of industrial centers in the United States, Europe and Japan; accelerated industrialization of several Third World nations; and increased internationalization of the financial industry into a global network of transactions (Sassen 1991). With the emerging spatial organization of the new international division of labor, John Friedmann identified a set of theses known as the “world city hypotheses” concerning the contradictory relations between production in the era of global management and political determination of territorial interests (Friedmann 1986).

Saskia Sassen and others have further elaborated the “global city hypotheses.” Global cities have acquired new functions beyond acting as centers of international trade and banking. They have become: (1) concentrated control locations in the world-economy that use advanced telecommunication facilities, (2) important centers for finance and specialized producer service firms, (3) coordinators of state power, (4) sites of innovative post-Fordist forms of industrialization and production, and (5) markets for the products and innovations produced (Sassen 2001a, 2000, 1991; Brenner 1998; Yeoh 1999; Hall 1996; Friedmann 1995). These structural shifts in the functioning of cities have “impacted both the international economic activity and urban form where major cities concentrate control over vast resources, while financial and specialized service industries have restructured the urban social and economic order” (Sassen 1991, pg 4). During the 1990’s, for example, New York specialized in equity trading, London in currency trading, and Tokyo in size of bank deposits (Slater 2003). Beaverstock, Smith and Taylor (1999) use Sassen’s focus on producer services to classify 55 cities as alpha, beta and/or gamma world cities based on the presence of accountancy, advertising, banking/finance and law firms. The “Globalization and World Cities Study Group and Network” at Loughborough University have developed a website ( that is a valuable resource for the study of systems of world cities

The most important assertion in the global cities literature is the idea that the global cities are cooperating with each other more than the world cities did in earlier periods. The most relevant earlier period is the Pax Britannica, especially the last decades of the nineteenth century. If this hypothesis is correct, the division of labor and institutionalized cooperative linkages between contemporary New York, London and Tokyo should be greater than were similar linkages between London, Paris, Berlin and New York in the nineteenth century. Obviously, communications technologies were not as developed in the nineteenth century, though intercontinental telegraph cables had already been laid, and Japan was not yet a core power in the world-system. But the nature and strength of coordination and cooperation among the world cities of the nineteenth century needs to be examined in order to support the hypothesis of greater contemporary integration that the global cities literature assumes.

Another important hypothesis of the global cities literature is based on Saskia Sassen’s (1991) observations about class polarization and the casualization of work within globalizing cities. The research of Gareth Stedman Jones on Irish immigration into London’s East End in the mid-nineteenth century (Jones n.d.) shows that a somewhat similar process of “peripheralization of the core” was occurring during the Pax Britannica. To be sure, much of the research on the global city system has been based on case studies of particular cities. Researchers generally seek to identify the processes leading to a specific city’s emergence and positioning within the larger system (Baum 1997; Grosfoguel 1995; Todd 1995; Machimura 1992; Kowarick and de Mello 1986). Janet Abu-Lughod (1999) traces the developmental histories of New York City, Chicago, and Los Angeles through their upward mobility in the world city system. While these U.S. metropoles share similar characteristics with other world cities, they have too many differences in geography, original economic functions, transportation, and political history to serve as much more than fascinating cases for comparative analyses of globalization. 

City Regions  

Text Box: Figure 1: City regions as indexed by light captured by satellite images. Another phenomenon of recent urbanization is the emergence of city-regions, large areas in which big cities are located rather closely to one-another and intervening areas are mainly suburbanized. Urban geographers have noted that populations in the rural areas and small towns of core countries are thinning and people are concentrating in these city regions (Scott 2001; Simmonds and Hack 2000). The city region phenomenon is made plain by examining Figure 1, a global map of city lights at night produced from satellite images.


All the continents have city regions, but the largest are those found in the eastern half of the United States and the western portion of Europe, with several other regions also displaying this phenomenon. These city regions are linked together spatially by overlapping surburban areas, and it is for that reason that our focus in this research is on the phenomenon of suburbanization. We will develop a method of spatially bounding multicentric city-regions that will enable is to quantitatively compare these with one another in terms of spatial and demographic sizes, population density, and settlement size distributions. We shall also study differences in their macrourban structures.



From the ancient world until the industrial age most cities had a monumental non-residential center surrounded by relatively high-density residential districts, with density largely limited by the technology available for creating structures either well below or well above ground level. Walled cities enclosed these high-density residences, but when the cities grew, suburban districts of rather lower densities formed outside of the old wall. The spatial extent of these cities was also limited by the fact that most traffic was pedestrian. This radioconcentric pattern continued to characterize cities in the industrial age despite the geometric decline of transportation costs produced by the steam-powered railways.  Improved transportation arose earliest and most quickly in the capitalist core cities and so their urban form was first affected by these developments, especially during the second half of the nineteenth century when the central portions of cities like London, New York, and Paris were rebuilt.  This was, however, only an interstitial period between the earlier pedestrian cities and the modern automobile-driven (literally) city.  New cities, such as Los Angeles, built during the automobile age from the early twentieth century to the present have been increasingly characterized by being multicentric and low-density (Dear 2000; 2002).  This also characterizes many of the new cities in non-core developing countries, which have also been based on automobile (or at least motorized bus) dependency.  And of course the low-density and multicentric pattern has been added to old concentric-style cities as they have experienced further growth during the automobile age.  This is accentuated by so-called edge cities (Garreau 1991).  Once again, these changes are most obvious in the core capitalist cities, but this urban form has diffused quickly to the non-core cities as they have been inundated with population increase.

There is a vast literature on the ecology or spatial form of urban places, but we distinguish three main types of modern urban macrostructure:

(1)   Type A: concentric-radial cities organized around a central business district with transportation corridors radiating out from it;

(2)   Type B: multicentric low density cities that are mainly “suburban” with relatively small non-residential centers dispersed across the built-up landscape (e.g. L.A.); and

(3)   Type C: a mixture of these two where the older concentric structure has become edged by a newer multicentric and low-density region.

We posit that, despite some gentrification processes, city growth in the future will largely occur in the suburban areas of cities, through the dual processes of suburban intensification (in-fill of existing suburban areas), and suburban extensification  (urban sprawl). These are the processes that lead to the creation of the edge cities that eventually become surrounded by their own suburbs and thus become new urban nuclei in the sea of suburbanization.  Therefore, if we are to understand the emerging global city system, we must have a good handle on the way in which urban places are evolving, especially in terms of suburbanization.


The Global City System

We want to study changes in the global city-size distribution because we are interested in the relationship between cities and power, and because the apparent flattening of the global city-size distribution discovered in the 1980s raises interesting questions about the upper limits of the sizes of megacities. Why did the global city-size distribution flatten out after 1950, modifying a pattern that had existed throughout the British and U.S. hegemonies in which the most powerful country had the largest city and there was a hierarchy of city population sizes revealed by the world’s largest cities (Chase-Dunn 1985)? Roland Fletcher  (personal communication) contends that contemporary institutional and infrastructural inventions only allow for megacities to function at maximum populations of around twenty millions and this serves as a kind of ceiling effect which has allowed cities in the non-core to catch up in terms of population size with the largest cities in the most powerful states. This may be what has produced the flat global city-size distribution that emerged after 1950. Fletcher’s notion of an upper limit on the sizes of large contiguous cities might also be part of the explanation for the emergence of city-regions rather than gigacities (the logical phase beyond megacities).


In order to study the global city-size distribution and the phenomenon of city-regions we are developing new methods of spatially bounding cities and city-regions using satellite data. Spatially bounding cities has long been problematic because information is often organized in terms of juridical boundaries.  We hypothesize that the overall rate of urban growth is correlated with economic growth at the level of the world-system as a whole. We expect to find faster rates of urban population growth during periods of faster economic growth, and that urban growth is cross-nationally correlated with economic growth. In peripheral regions the relationship with economic growth may be reduced because migration to cities is driven by the redundancy of rural populations. Thus at the level of the whole world-system we predict that the rate of urban growth declined after 1980 as the world economy moved into stagnation. In order to investigate this we will use census data since 1950 on the world largest cities and data on GDP growth on the countries in which these cities are located.

      We also hypothesize that the rate of suburbanization is related to overall economic growth and to the level of development in the world economy. We expect slower and less suburbanization in less developed countries and fewer cities of the Los Angeles type. Our suppositions here are based on the differences in income and the affordability of automobile transportation to poor families. We would expect that suburbanization in semiperipheral urban regions such as Mexico City is higher density because poor urban residents are unable to purchase their own vehicles and are unable to purchase large residential lots.

      We also hypothesize that the relative investments in mass transportation compared to road-building affect the rate of suburbanization. Here we must control for the age of existing urban infrastructure because it is much more expensive to rebuild than to build on vacant land. We can estimate the historical age of our cities, that date at which they reached population sizes of 100,000, and use this as a control in our analysis of the effects of mass transportation investment and road building.

      We also suppose that the size and density of city-regions are related to global differences in the level of development, and that once these features of city regions are taken into account it will turn out that the global city-region-size hierarchy is indeed related to economic and political/military power as it has been in the past. Developing countries have succeeded in building very large megacities, but their city-regions are not as large and dense as those in the core. Thus once we get the unit of analysis right--city-regions rather than single urban agglomerations—we are likely to find that the old association between power and settlement size continues to hold in the modern world-system.

Measuring the growth of suburbanization requires standardized categories of population density that are comparable across cities in different regions, cultures and levels of national development.  We propose to use remotely sensed imagery to classify land cover according to an urban gradient that distinguishes the mix of land cover associated with urban, suburban, and rural land uses.  We will calibrate these measures to census and ancillary land-use data from administrative sources in order to create replicable algorithms that can be used globally to describe and model changes in urban macrostructures and thus in the global city system.  Part of the calibration will involve the estimation of population size (and thus density) from the satellite imagery. Measurement of the population sizes of cities is not without difficulties. How can we know the number of people who reside in Los Angeles today? We use the most recent census, a survey of  “residents” conducted by the U.S. federal government. What are the spatial boundaries of “Los Angeles”? Do we mean the city of Los Angeles, Los Angeles County, the contiguous built-up area that constitutes “greater Los Angeles,” or a definition based on the proportion of the local population that is employed in “Los Angeles”? Does “Los Angeles” include San Diego? Nighttime satellite photos of city lights imply a single unbroken megalopolis from Santa Barbara to Tijuana:  So where is Los Angeles? We use the contiguous built-up area as our main way of spatially defining cities. Urban geographers have made considerable progress on the task of using satellite data to spatially bound cities (Weber 2001).

Our work has clear significance for the testing of theories about city-regions and the way in which they contribute to the overall world-system that has emerged since the creation of capitalist cities several centuries ago.  It also has significance for the increasingly fruitful ways in which remotely sensed imagery has become a powerful adjunct to the methods used in the social sciences.  Higher spatial resolution imagery that covers the entire globe with increasing temporal resolution, combined with emerging methods of quantifying those images in ways that make sense of the urban scene, have allowed us to think about testing theories in ways that were previously unimaginable.  Thus, the results from this research will contribute to a growing linkage between methods originated in the physical and natural sciences and theories developed in the social sciences.  This offers a potentially new outlook as researchers are better able to delineate the intertwining of the physical and social worlds.


The Research Team  

This project builds on the NSF-funded work of both of the Co-PIs, and the project is staffed by a set of researchers who have substantial expertise in all aspects of the work that we are developing.  Christopher Chase-Dunn is Distinguished Professor of Sociology and Director of the Institute for Research on World-Systems at the University of California-Riverside. He received his Ph.D in Sociology from Stanford University in 1975. Chase-Dunn has done crossnational quantitative studies on the effects of dependence on foreign investment. His recent research focuses on intersocietal systems, including both the modern global political economy and earlier regional world-systems. He is doing research on the causes of empire expansion and urban growth (and decline) in the Afroeurasian world-system over the last 4000 years. Chase-Dunn is the founder and co-editor of the electronic Journal of World-Systems Research and the Series Editor of two book series on global social change published by the Johns Hopkins University Press.   In 2001 he was elected a Fellow of the American Association for the Advancement of Science. In 2002 he was elected President of the Research Committee on Economy and Society (RC02) of the International Sociological Association.

John R. Weeks is Professor of Geography and Director of the International Population Center at San Diego State University (SDSU).  He also holds an appointment as Clinical Professor of Family and Preventive Medicine at the University of California, San Diego, School of Medicine.  As noted above, he is currently the Principal Investigator on a project funded by the National Science Foundation (BCS-0095641) to apply remotely sensed imagery and GIS to an analysis of the Arab Fertility Transition, focusing on Egypt and Jordan.  He received his Ph.D. in Demography in 1972 from the University of California, Berkeley, where his mentor was the eminent sociologist and demographer Kingsley Davis, one of the most important analysts of world urbanization. Davis’s work also greatly influenced Chase-Dunn research on city systems. Dr. Weeks will be linked to the project through a contractual arrangement with San Diego State University.  The project will permit expertise to be gained in these areas of research by a graduate student and an undergraduate student at the University of California, Riverside, and by two graduate students at San Diego State University.



Our work will focus on eight tasks that will proceed in tandem, with the results coming together in March of 2006: (1) quantification of data for the entire population of world cities, to be used to test theories about city systems; (2) classification of satellite imagery for five cities for three dates each; (3) estimation of measurement error due to satellite imagery; (4) linking of census data and, where available, ancillary land use data with the data from the satellite imagery; (5) calibration of the census and land use data with the census land cover classification variables; (6) estimation of “real” change over time and space in the spatial extent and population density of each of the study sites; (7) collation of all results in order to test theories about the role of suburbanization and related urban processes on change occurring in core and peripheral cities; (8) dissemination of results and establishment of plans for Phase II of the overall project.


Data for all World Cities

 We have identified 251 cities as the largest on Earth in 2000. These cities on all continents and in Oceania serve as the main focus of our overall project. We will also study the ten largest city-regions, which include: Southern California/Northwestern Mexico, Midwestern-Eastern United States and adjacent Canada, Central Mexico, Northwestern South America, the River Plate urban region, Southeastern Brazil, Southeastern South Africa, Europe-North Africa-Western Asia, South Asia, China and Japan. The project will build a data set with quantitative measures on 251 cities, all the countries in which these cities are located, ten city-regions, and variable characteristics of the world-system as a whole. These data sets will be used to test the hypotheses (above) about the causes of different kinds of urban growth. This phase of our study focuses mainly on the period between 1984 and 2005, though we will also use census data since 1950 and some remote sensing data before 1984. The recent time period of 1984 to 2005 is dictated largely by the availability of suitable satellite imagery. LANDSAT imagery prior to 1982 was at 79-meter resolution, which we judge to be too coarse for meaningful urban analysis. The LANDSAT TM, with a higher spatial resolution of 30 meters, went up in 1982, but the imagery archive does not go back consistently before 1984.  Every city is covered from the mid-1980s to the present, so it will be possible to choose the best scene for the specific date (looking for imagery taken at the same time of year, so that seasonal vegetation doesn't confound the analysis). We are also collaborating with the Urban Environmental Monitoring project (UEM 2000) and will make use of the ASTER results of that study.


Study Sites for data from Satellite Imagery 

In order to evaluate the utility of using remotely sensed data for the study of city-regions, we will focus on five targeted cities: Los Angeles (USA), San Diego (USA), Cairo (Egypt), Amman (Jordan) and Accra (Ghana). These targeted cities will be used to develop our mulitiple indicator measurement error model (see below) for sorting out apparent temporal differences that are due to changes in remote sensing technology rather than due to real changes in the cities we are studying. These cities are both similar and different in ways that will be helpful in building our measurement model. Los Angeles, San Diego, Cairo and Amman are in semiarid and Mediterranean climates, whereas Accra is in a tropical climate. Los Angeles and San Diego are recently built low-density, multicentric cities (Type B), whereas Cairo, Amman and Accra are concentric/radial older cities with newer suburban fringes (Type C).   These are cities for which Dr. Weeks already has some of the imagery needed for this study.  For Los Angeles, Cairo and Accra, we are leveraging work carried out by Weeks and his associates in two previous NSF-funded projects--Grant BCS-0095641 (which was discussed above), and BCS-0117863 (Doctoral Dissertation Research: Environmental Context of Social Vulnerability to Urban Earthquake Hazards, funded from 8/1/01 through 7/31/03, which supported the dissertation research of Dr. Tarek Rashed, now Assistant Professor of Geography at the University of Oklahoma).  Thus, we are able to leverage nearly $30,000 worth of multi-spectral and panchromatic, high and medium spatial resolution imagery for this project by focusing on those three cities.  From other sources, Dr. Weeks already has some of the imagery required for San Diego and Accra.  Thus, the selection of these cities is grounded both on the basis of a good comparative fit, and also on the ability of the researchers to accomplish more than would otherwise be possible, by building on work already supported by NSF.


Classification of the Remotely Sensed Imagery 

For each of the five cities, we propose to use Landsat TM imagery for 1984 (the first year for which TM are data for all five cities), 1994, and 2004 (available at about the time of funding of this project).  This will provide a consistent set of data from the same type of platform and sensor.  The principle issue with the TM data is that with a spatial resolution of 30m the accuracy of the land cover classification in urban areas can be problematic.  We have dealt with this problem in our current research by using “soft” classification methods, rather than “hard” methods, as discussed below.  We are also interested, however, in the extent to which we may learn more about urban places by using higher resolution imagery.  Thus, for each of the five cities we propose to compare the 2004 TM classification results with the results of classifying ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer) imagery, which has three spectral bands in the visible near-infrared (VNIR) range at a spectral resolution of 15m.  Additionally, for San Diego, Cairo, and Accra, we will be able to compare the TM and ASTER results with results from much higher resolution Ikonos (5m multispectral, available for San Diego), and Quickbird (2.4m multispectral, available for Cairo and Accra).  These data, which are already in the SDSU archive, obviously refer to dates prior to 2004 (2000-2002), but that small difference in temporal resolution should not be problematic for our analysis.

In order to appreciate the value of remotely sensed imagery for analysis of urban places, it is crucial to understand exactly what information can be extracted from such images.  The image itself is composed of a two-dimensional array of pixels from which radiant energy has been captured for an area on the ground that is equal to the spatial resolution of the image.  The information recorded for each image depends upon the particular sensor, but the brightness within a given band is assigned a digital number.  The combination of digital numbers representing relative reflectance across the different bands of light yields the spectral signature of that pixel.  Particular types of land cover (e.g, vegetation, soil, water, impervious surface) tend to have unique spectral signatures. The more bands that a sensor has the more detailed can be the land cover classification.  If there are only a few bands, it is possible to differentiate vegetation from non-vegetation, but with more bands it may be possible to differentiate a field of corn from a field of wheat or, within the urban area it may be possible to differentiate a tin roof from a tile roof.  The typical tradeoff in imagery is that lower spatial resolution imagery will tend to have more bands (i.e., higher spectral resolution) than higher spatial resolution imagery.  Our team’s experience working with imagery for urban places suggests thus far that higher spatial resolution is more important in characterizing an urban place than is the number of bands available for analysis (Rashed and Weeks 2003; Rashed et al. 2000; Rashed et al. 2001; Weeks 2003c), and Aplin (2003) has reported similar conclusions. This is because the built environment is, obviously, configured differently than the natural environment, and the two most useful ways that we have found of quantifying urban places from imagery are in terms of (1) the proportional abundance or composition of fundamental land cover classes; and (2) the spatial configuration of the pixels identified with each land cover class.

A common first task in using the data recorded for each pixel is to determine what type of land cover is represented by that pixel.  Do the data represent vegetation (and perhaps a specific type of vegetation), or bare soil, water, shade, or an impervious surface (such as the roofing material of a building or the asphalt or cement of roads)?  These surface materials are the basic building blocks of natural and built environments and each type of land cover is associated with a particular spectral signature. The higher the spatial resolution the more accurately a pixel can be classified into basic land cover types because it is more likely that the pixel will include only one type of land cover.  On the other hand, for lower resolution images, the more likely it is that the pixel will represent a mixture of different land covers, forcing a decision about how to appropriately classify the image. 

In a “hard” classification, each pixel in an image is assigned to one land cover class, using one of several statistical algorithms to determine the final choice.  However, in all but very high-resolution imagery, the area represented in a pixel from an urban scene is likely to be a mix of land covers.  For this reason, we have favored a classification scheme known as spectral mixture analysis (SMA), which “unmixes” each pixel into its constituent land cover classes and assigns a percentage of land cover class to each pixel. This is often called a “soft” classification approach because each pixel is described not in terms of a single land cover, but rather in terms of the proportional abundance (fraction) of each land cover class.  The SMA approach assumes that a landscape is formed from continuously varying proportions of idealized types of land cover with pure spectra, called endmembers (Adams, Smith, and Gillespie 1993). Endmembers are abstractions of land cover materials with uniform properties present in the scene. In an urban environment, these may include impervious surfaces, vegetation types, water bodies, and bare soils (Ridd, 1995).

Linear SMA is the process of solving for endmember fractions, assuming that the spectrum measured for each pixel represents a linear combination of endmember spectra that corresponds to the physical mixture of surface components weighted by their areal abundance. However, if a pixel is modeled by fewer endmembers than required, the unmodeled portion of the pixel spectrum will be partitioned into the resultant fractions, thus increasing the model error for that pixel (Roberts et al. 1998).  To correct for this problem, we have successfully tested the applicability of an algorithm utilizing the technique of multiple endmember spectral mixture analysis (MESMA) to measure the physical composition of urban morphology from a Landsat Thematic Mapper (TM) multispectral image (Rashed et al. 2003 (forthcoming)-a). It has been suggested that urban morphology is “the physical appearance of social reality”(Pesaresi and Bianchin 2001:56). The potential of MESMA to contribute to urban morphological analysis lies in its ability to quantify the physical composition of urban areas occasioned by human activity at different geographic scales.

In classifying the data by land cover class, we have previously employed Ridd’s (1995) V-I-S (vegetation, impervious surface, soil) model to guide a spectral mixture analysis of medium-to-high resolution multi-spectral images for Cairo for 1986 and 1996, in a manner similar to methods used by Phinn and his colleagues for Brisbane, Australia (Phinn et al. 2002), and by Wu and Murray (2003) for Columbus, Ohio.  The V-I-S model views the urban scene as being composed of combinations of three distinct land cover classes.  An area that is composed entirely of bare soil would be characteristic of desert wilderness, whereas an area composed entirely of vegetation would be dense forest, lawn, or intensive fields of crops.  At the top of the pyramid is impervious surface, an abundance of which is characteristic of central business districts, which are conceptualized as the most urban of the built environments.

We have added another component to Ridd’s physical model—shade/water—following the work of Ward, Phinn, and Murray (2000) suggesting that the fourth physical component improves the model in settings outside of the United States.  When combined with impervious surfaces in urban areas it becomes a measure of the presence of multi-story buildings (based on the shadows cast by buildings).  When combined with vegetation it provides a measure of the amount of water in the soil and the shade cast by tall vegetation (largely trees that may serve as windbreaks in agricultural areas).  In combination with bare soil it is largely a measure of any shadows cast by trees, although there could be some component of shade from large buildings in heavy industrial areas.

Once an image is classified according to land cover types, information from other sources may be used to make inferences about the way in which the land is being used, since land use is a socially derived category.  From this process, variables may be created that describe the environmental context of a specific place.  Thus, spatial aggregation of the land cover data for all pixels in an area (such as a census tract) yields a measure of the area’s land cover composition.  To these we add algorithms for quantifying the spatial configuration of the pixels of specific land cover classes (known as “patches”) in a given area (such as a census tract) (McGarigal 2002). These landscape metrics were developed originally for applications in landscape ecology, but have recently been discovered to have considerable potential value for describing the urban environment (Herold, Scepan, and Clarke 2002).

The landscape metric algorithms allow us to produce several indices of the way in which each land cover class is organized spatially.  These include, in particular, shape complexity and isolation/contiguity of class types, based on concepts of fractal geometry as applied to geography (see, for example, Lam and De Cola 1993). Work is only now beginning on the creation of what might be thought of as a “reference library” of landscape metrics that are consistently related to particular kinds of urban places.  Indeed, the study by Herold, Scepan and Clarke (2002) is one of the very first of its kind.  This is, of course, one of the ways in which the proposed research is moving into quite literally uncharted territory, but the potential is very high that these measures will allow us to create much more meaningful quantifiable indices of the built and natural environments related to the urbanness. For example, the proportional abundance of vegetation, as classified from the imagery, tells us whether there is any vegetation within a given urban area.  The contagion index of landscape metrics can then tell us how clumped together the patches of vegetation are.  If this index is high, we can expect that the vegetation represents a true green space -- perhaps a park-- rather than randomly placed plants. The high-resolution panchromatic imagery will allow us to confirm whether our interpretation of the landscape metric appears to be correct. As another example, two neighborhoods may have identical proportional abundances of impervious surface, reflecting (literally) the roofs of buildings.  The landscape metric that measures the perimeter-area ratio then tells us how those pixels are arranged.  If the ratio is very low, then the image has probably captured a large building (which is probably connected to local infrastructure), whereas if the ratio is high, the image has captured a number of different buildings, implying that as the ratio increases, the number of individual buildings is going up.  If there are a number of small buildings in a neighborhood with little vegetation and perhaps some bare soil, then we may infer that the neighborhood can be characterized as a low-rise slum (with small homes that are probably not connected to infrastructure).  If the ratio is lower and the proportional abundance of vegetation is higher, then we may infer that the neighborhood is a more prosperous residential area.  As the perimeter-area ratio increases with the same amount of vegetation, we may infer that the size of the buildings is increasing, perhaps indicating higher-status villas.  Once again, these interpretations can be checked by reference to the high-resolution panchromatic imagery. Members of the research team already have experience utilizing the Fragstats program ( working with remotely sensed data for Cairo, Egypt and Amman to establish a baseline of metrics for use with the data for the five study sites selected for this research.


Measurement Error  

We will develop multiple indicator measurement error models of the spatial and demographic features of the world largest cities and city-regions in order to quantify spatial and population growth and patterns of suburbanization from 1984 to 2004. For example, Daniel Pasciuti has developed a measurement error model approach to estimating the population sizes of preindustrial cities from a large set of proxy indicators (see  We will apply this same approach to the technology change measurement problem. Multiple indicator measurement error models utilize structural equations modeling to estimate the relationships among a set of proxy indicators of an underlying (latent) variable characteristic of a substantial number of cases (Bollen 1989).  We will develop these measurement models in order resolve the problem of measurement error due to changing satellite imagery technology over the period we wish to study.  These methods, derived from social science research, will then be evaluated in comparison with more traditional physical science based measures of RS measurement error.

This problem is closely analogous to other efforts to study change over time in which earlier, less complete and more error-prone data are used in conjunction with later, more complete and more accurate data. The problem is to sort out the real change from the apparent changes that are due to early measurement error. The structural equations approach allows us to estimate the sizes of errors in the earlier estimates and to improve those estimates based on measurement error models constructed using the later and better instruments. In time-series analysis it is common to exploit overlaps between different measures to understand the measurement differences. This is what we will be doing when we build a measurement error model that estimates the relationships among indicators, and then uses these estimates to correct earlier estimates for which we do not have all the information.

Linking RS data with Census and Other Data 



Once the RS data are classified, the land cover data will be aggregated at the census block-group level for each city for each date under consideration, landscape metrics will be calculated, and an urban gradient score will be calculated.  This will build upon the work already underway by Weeks and his associates (see, for example, Weeks 2003c; Weeks et al. 2003).  The idea is to create an urban gradient index from the RS data and then calibrate that index to the census and other data available for that census block-group.  The basic idea is illustrated in Figure 3 below. City and urban region characteristics (area size, population size, urban structural type, population densities, etc. will be estimated using geocoded census data (ground truth), high resolution remote sensing imagery, LANDSAT (TM), ASTER imagery and other imagery. The models that we are able to build using our five target cities will be rough estimates that will be refined in the second stage of our study when we turn to the analysis of all 251 largest cities. For most of these we will have only general estimates based on census data (e.g. total population, total land area). We will have both ASTER and TM imagery after 2001, and so will be able to test and improve our model using these cases.


Text Box: Figure 3: Multiple Indicator Measurement Error Model of City Characteristics





As long ago as 1980, Clayton and Estes (1980), building on the pioneering work by Tobler (1969), demonstrated the close relationship in the United States between high resolution remotely sensed imagery and census data.  Lo (1995) showed that high-resolution multispectral SPOT imagery could be used to model population and dwelling unit estimation in the Hong Kong metropolitan area, although he found that at the micro level of spatial scale the results were less accurate than for larger areas. However, improved estimating techniques led subsequently to acceptable results at the census tract level in Atlanta (Lo, 2003). Sutton and his associates (Sutton et al. 1997, 2001; Sutton 2003) have shown that night-time light imagery (see Figure 1 above) is correlated with population density, but the spatial resolution (one kilometer) of these results is coarser than we believe is required in order to adequately model the urban processes of suburbanization.  Our method uses higher resolution imagery and incorporates measures not only of land cover abundance, but also land cover configuration.  Thus, we have a relatively complex set of RS measures with which to associate population data at the zonal level (typically the census tract). This linkage will be accomplished using regression methods, following the lead of other researchers (see Harvey 2003 for a review), but we will then use dasymetric mapping techniques (see, for example, Yuan, Smith, and Limp 1997; and Weeks et al. 2000) and allometric growth modeling techniques (see Longley and Mesev 2001; and Lo 2003) to fine-tune the estimates to a grid that will be larger than a single pixel, but substantially less than one kilometer.

We will use structural equations modeling and time-series analyses to test the causal propositions delineated above at three levels of analysis: cities, city-regions and the world-system as a whole. We will also use our geocoded data to build a Time-mapped GIS for scientific visualization of urban growth processes and for our educational World Cities web presentation. The TimeMap® Project ( is a temporal geographical information system (TGIS) that utilizes standardized and interoperable web-enabled datasets to produce animated maps that show change over time (Johnson 2000).


This proposed research project is developing new methods for using remotely sensed data from satellites to study the growth of world cities and city-regions, the rates of suburbanization and trends in the global size distribution of cities.  Whether or not rates of growth and suburbanization are accelerating, steady, or decreasing, and how these may vary between core, peripheral and semiperipheral societies in the world-system, will have important implications for theories of urban growth and for the future of sustainability.


References Cited

Abu-Lughod, Janet L. 1999. New York, Chicago, Los Angeles: America’s Global Cities. Minneapolis, Minnesota: University of Minnesota Press.

Adams, J.B., M.O. Smith, and A.R. Gillespie. 1993. "Imaging Spectroscopy: Interpretaton Based on Spectral Mixture Analysis." in Remote Geochemical Analysis: Elements and Mineralogical Composition, edited by C. M. Pieters and A. J. Englert. Cambridge, UK: Cambridge University Press.

Aplin, Paul. 2003. “Comparison of simulated IKONOS and SPOT HRV imagery for classifying urban areas,” pp: 23-45 in Victor Mesev, (ed.), Remotely Sensed Cities. London: Taylor & Francis.

Arrighi, Giovanni 1994 The Long Twentieth Century London: Verso.

Baum, Scott. 1997. “Sydney, Australia: a global city? Testing the social polarization thesis.”  Urban Studies 34, 11: 1881-1901.

Beaverstock, J.V., P.J. Taylor and R.G. Smith. 1999. “A roster of world cities.” Cities 16: 445-458.

Bollen, Kenneth A. 1989 Structural Equations with Latent Variables. New York: John Wiley.

Bornschier, Volker and Christopher Chase-Dunn (eds.)  2000 The Future of Global Conflict. London: Sage

Bosworth, Andrew. 2000. “The Evolution of the World-City System, 3000 BCE to AD 2000.” In World System History: The Social Science of Long-Term Change, ed. Robert A. Denmark, Jonathan Friedman, Barry K. Gills, and George Modelski. New York: Routledge.

Boulding, Kenneth. 1978. “The city as an element in the international system.” In Systems of Cities, ed. L.S. Bourne and J.W. Simmons. New York: Oxford University Press.

Brenner, Neil. 1998. “Global Cities, Global States: Global City Formation and State Territorial Restructuring in Contemporary Europe.” Review Of International Political Economy 5, 1: 1-37.

Brivio, Pietro A. and Eugenio Zilioli (2001) “Urban pattern characterization through geostatistical analysis of satellite images.” Pp. 39-53 in Donnay, et al.

Castells, Manuel 1989 The Informational City. Cambridge, MA: Blackwell

Chase-Dunn, Christopher. 1982. “World division of labor and the development of city systems.” Comparative Research 9, 3: 3-9.

Chase-Dunn, Christopher. 1985a. “The System of World Cities, A.D. 800- 1975.”        In Urbanization And The World-Economy, ed. Michael Timberlake. New York: Academic Press. 

Chase-Dunn, Christopher. 1985b. “The coming of urban primacy in Latin America.” Comparative Urban Research 11, 1-2: 14-31.

Chase-Dunn, Christopher. 1992. “The changing role of cities in world-systems.” In World Society Studies, ed. Volker Bornschier and Peter Lengyel. Frankfurt and New York: Campus Verlag.

Christopher Chase-Dunn, Yukio Kawano and Benjamin Brewer 2000 "Trade Globalization since 1795: waves of integration in the world-system," American Sociological Review 65:77-95.

Christopher Chase-Dunn and Andrew Jorgenson, "Systems of Cities,"  in Paul Demeny and Geoffrey McNicoll. 2003. Encyclopedia of Population. New York: Macmillan Reference USA.

Chase-Dunn, Christopher and Alice Willard. 1994. “Cities in the central political-military network since CE 1200.” Comparative Civilizations Review 30:104-32 (Spring).

Chen, Xiangming. 1995. “Chicago as a global city.” Chicago Office 5: 15-20.

Clayton, Christopher and John E. Estes.  1980. “Image Analysis as a Check on Census Enumeration Accuracy.” Photogrammetric Engineering and Remote Sensing 46(6):757-764.

Cronon, William 1991 Chicago and the Great West. New York: Norton.

Davis, Kingsley. 1972. World Urbanization 1950-1970, Volume 2: Analysis of Trends, Relationships, and Development. Berkeley, CA: Institute of International Studies, University of California.

—. 1973. Cities and Mortality. Ličge: IUSSP.

Davis, Mike 1990 City of Quartz, London: Verso

Dear, Michael J. 2000 The Postmodern Urban Condition. Malden, MA: Blackwell.

Dear, Michael  J. (ed.) 2002 From Chicago to L.A.. Thousand Oaks, CA: Sage.

Donnay, Jean-Paul, Mike J. Barnsley and Paul A. Longley (eds.) Remote Sensing and Urban Analysis. New York: Taylor and Francis.

Fletcher, Roland 1995 The Limits of Settlement Growth. Cambridge:  Cambridge University Press.

Friedmann, John. 1986. “The World City Hypothesis.” Development and Change 17, 1: 69-84.

Friedmann, John. 1995. “Where We Stand: A Decade of World City Research.” In World Cities in a World-System, ed. Paul Knox and Peter Taylor. New York: Cambridge University Press.

Fugate, Debbie. 2003. "The use of remote sensing, GIS, and spatial statistics to examine the relationship between environmental context and health in Cairo, Egypt." M.A. Thesis, Department of Geography, San Diego State University.

Garreau, Joel. 1991. Edge City: Life on the New Frontier. New York: Doubleday.

Godfrey, B.J. and Y. Zhou. 1999. “Ranking world cities: multinational corporations and the global urban hierarchy.” Urban Geography 20: 268-281.

Gottmann, J. 1989. “What are cities becoming the centres of? Sorting out the possibilities.” In Cities in a Global Society, ed. R.V. Knight and G. Gappert. London: Sage.

Grosfoguel, Ramon. 1994. “World cities in the Caribbean: the rise of Miami and San Juan.” Review 17, 3: 351-381.

Hall, Peter. 1996. “The Global City.” International Social Science Journal 48, 1: 15-23.

Hall, Peter. 1998. “Globalization and the world of cities.” In Globalization and the World of Large Cities, ed. F-c Lo and Y-m Yeung. Tokyo: United Nations University Press.

Harvey, Jack T. 2003.  “Population estimation at the pixel level,” pp. 181-205 in Victor Mesev, (ed.), Remotely Sensed Cities. London: Taylor & Francis.

Inoguchi, Takashi ,Edward Newman, and Glen Paoletto (eds.) 1999 Cities and the environment: new approaches for eco-societies. New York : United Nations University Press.

Hepner, G.F., B. Houshmand, I. Kolikov and N. Bryant, "Investigation of the Integration of AVIRIS and IFSAR for Urban Analysis," Photogrammetric Engineering and Remote Sensing, 64-8 (August, 1998),pp 813-820.

Herold, Martin, Joseph Scepan, and Keith C. Clarke. 2002. "The use of remote sensing and landscape metrics to describe structures and changes in urban land uses." Environment and Planning A 34:1443-1458.

Johnson, Ian 2000 “A step-by-step guide to setting up a TimeMap dataset.” Archaeological Computing Laboratory. University of Sydney.

Jones, Gareth Stedman. No date.  A Study in the Relationship Between Classes in Victorian Society.  Part 1: The London Labour Market and the Causal Labour Problem.  London: Penguin.  (ISBN 0 14 055. 103 4)

King, Anthony D. King. 1990. Global Cities: Post-Imperialism and the Internationalism of London. London: Routledge.

Kowarick, L. and M. Campanario. 1986. “San Paulo: the price of world city status.”  Development and Change 17, 1: 159-174.

Lam, Nina and Lee De Cola. 1993. "Fractals in Geography.". Englewood Cliffs, NJ: Prentice-Hall.

Lo, Chor Pang 1995. Automated population and dwelling unit estimation from high-resolution satellite images: A GIS approach.” International Journal of Remote Sensing 16(1):17-34.

__. 2003. “Zone-based estimation of population and housing unites from satellite-generated land use/land cover maps,” pp: 157-180 in Victor Mesev, (ed.), Remotely Sensed Cities. London: Taylor & Francis.

Longley, Paul A., and Victor Mesev. 2001. “Measuring Urban Morphology using Remotely-Sensed Imagery,” pp. 163-183 in Jean-Paul Donnay, Mike J. Barnsley, and Paul A. Longley, (eds.), Remote Sensing and Urban Analysis.  London: Taylor & Francis.

Lyons, Donald and Scott Salmon. 1995. “World cities, multinational corporations,       and urban hierarchy: the case of the United States.” In World Cities in a World-System, ed. Paul Knox and Peter Taylor. New York: Cambridge University Press.

Marcuse, Peter and Ronald van Kempen (eds.) 2000 Globalizing cities : a new spatial order Malden, MA. : Blackwell.

Marshall, Alex, 2000  How cities work : suburbs, sprawl, and the roads not taken.   Austin:   University of Texas Press.

Meyer, D.R.. 1998. “World cities as financial centres.” In Globalization and the World of Large Cities, ed. F-c Lo and Y-m Yeung. Tokyo: United Nations University Press.

McGarigal, K. 2002. "FRAGSTATS Documentation, Background Material.".

Machimura, Takashi. 1992. “The urban restructuring process in the 1980’s: transforming Tokyo into a world city.” International Journal of Urban and Regional Research 16, 1: 114-128.

Meyer, David. 1986. “The world system of cities: relations between international financial metropolises and South American cities.” Social Forces 64: 553-81 (March).

Netzband, M. and W.I. Stefanov 2003 “Assessment of urban spatial variation using ASTER data,” Presented at the urban remote sensing conference in Regensburg.

Pasciuti, Daniel 2003 “A measurement error model for estimating the populations of cities,”

Pesaresi, M. and A. Bianchin. 2001. "Recognizing Settlement Structure using Mathematical Morphology and Image Texture." in Remote Sensing and Urban Analysis, edited by J.-P. Donnay, M. J. Barnsley, and P. A. Longley. London: Taylor & Francis.

Phinn, S.R., M. Stanford, P. Scarth, A.T. Murray, and P.T. Shyy. 2002. "Monitoring the composition of urban environments based on the vegetation-impervious surface-soil (VIS) model by subpixel analysis techniques." International Journal of Remote Sensing 23:4131-4153.

Ramsey, Michael 2003 “”Mapping the sity landscape from space: the ASTER Urban Environmental Monitoring Program.” In Earth Sciences in the Cities. American Geophysical Union.

Rashed, Tarek and John R. Weeks. 2003. "Assessing Vulnerability to Earthquake Hazards Through Spatial Multicriteria Analysis of Urban Areas." International Journal of Geographical Information Science 17:549-578.

Rashed, Tarek, John R. Weeks, M. Saad Gadalla, and Allan G. Hill. 2000. "Demographic Analysis Using Spectral Mixture Modeling of Remote Imagery for Urban Areas: A Case Study of Cairo, Egypt." in Annual Meeting of the Association of American Geographers. Pittsburgh.

—. 2001. "Revealing the Anatomy of Cities Through Spectral Mixture Analysis of Multispectral Imagery: A Case Study of the Greater Cairo Region, Egypt." Geocarto International 16:5-16.

Rashed, Tarek, John R. Weeks, Dar A. Roberts, John Rogan, and Rebecca Powell. 2003 (forthcoming)-a. "Measuring the Physical Composition of Urban Morphology using Multiple Endmember Spectral Mixture Analysis." Photogrammetric Engineering and Remote Sensing.

Rashed, Tarek, John R. Weeks, Douglas A. Stow, and Debbie Fugate. 2003 (forthcoming)-b. "Measuring Temporal Compositions of Urban Morphology through Spectral Mixture Analysis: Toward a Soft Approach to Change Analysis in Crowded Cities." International Journal of Remote Sensing.

Ridd, Meredith K. 1995 “Exploring a Vegetation-Impervious Surface-soil (VIS) model for urban ecosystems analysis through remote sensing: comparative anatomy for cities.” 1995  International Journal of Remote Sensing 16,12:2165-2186.

Roberts, D.A., M. Gardner, R. Church, S. Ustin, G. Scheer, and R.O. Green. 1998. "Mapping Chaparral in the Santa Monica Mountains Using Multiple Endmember Spectral Mixture models." Remote Sensing of the Environment 65:267-279.

Sassen, Saskia. 1991 and 2001 (second edition). The Global City: New York, London, Tokyo. Princeton, New Jersey: Princeton University Press.

Sassen, Saskia. 2000. Cities In A World Economy. Thousand Oaks, CA: Pine Forge Press.

Sassen, Saskia  2001. “Cities in the global economy.” In Handbook of Urban Studies, ed. Ronan Paddison. Thousand Oaks, California: Sage Publications.

Scott, Allen J. (ed.) 2001 Global City-Regions : trends, theory, policy  New York :             Oxford University Press

Sheehan, Molly O'Meara, Jane Peterson  (eds.) 2001 City limits : putting the brakes on sprawl . Washington, D.C. : Worldwatch Institute.

Simmonds, Roger and Gary Hack. (eds.) 2000 Global city regions : their emerging forms London : Spon.

Slater, Eric 2003 “The Flickering Global City: London-New York-Tokyo and the Hegemonic Transition”  Paper presented at the annual meeting on the Political Economy of World-Systems PEWS XXVII, Georgetown University, April 25-26, 2003

Smith, David A. 1996. Third World Cities in Global Perspective: The Political Economy of Uneven Urbanization. Boulder, Colorado: Westview Press.

Smith, David A. 2000. “Urbanization in the World-System: A Retrospective and Prospective.” In A World-Systems Reader: New Perspectives On Gender, Urbanism, Cultures, Indigenous Peoples, And Ecology, Thomas D. Hall (ed.). Boulder, Colorado: Rowman & Littlefield Publishers.

Smith, David A. and Michael Timberlake. 1993. “World cities: a political economy/global network approach.” Research in Urban Sociology 3: 181-207.

Smith, David A. and Michael Timberlake. 1995. “Cities in Global Matrices: Toward Mapping the World-System’s City System.” In World Cities in a World-System, ed. Paul L. Knox and Peter J. Taylor. New York, New York: Cambridge University Press.

Smith, David A. and Michael Timberlake. 1995. “Conceptualizing and mapping the structure of the world’s city system.” Urban Studies 32: 287-302.

Smith, David A. and Michael Timberlake. 1998. “Cities and the Spatial Articulation of the World Economy through Air Travel.” In Space And Transport In The World-System, Paul Ciccantell and Stephen G. Bunker (eds.). Westport, CT: Greenwood Press.

Smith, David A. and Michael Timberlake. 2001. “World City Networks and Hierarchies, 1977-1997: An Empirical Analysis of Global Air Travel Links.” American Behavioral Scientist 44, 10: 1656-1678.

Sutton, Paul. 2003. “Estimation of human population parameters using night-time satellite imagery,’ pp: 301-333 in Victor Mesev, (ed.), Remotely Sensed Cities. London: Taylor & Francis.

Sutton, Paul, Dar Roberts, Chris Elvidge, and Henk Meij. 1997. “A Comparison of Nighttime Satellite Imagery and Population Density for the Continental United States.” Photogrammetric Engineering and Remote Sensing 63(11):1303-1313.

Sutton, Paul, Dar Roberts, Chris Elvidge, and Kimberly Baugh. 2001. “Census from Heaven: an estimate of the global human population using night-time satellite imagery.” International Journal of Remote Sensing  22(16):3061-3076.

Taagepera, Rein and Edgar Kaskla. 2001. “The city-country rule: an extension of          the rank size-rule.” Journal of World-Systems Research 7: 157-174 (Fall).

Timberlake, Michael. 1985. “The world-system perspective and urbanization.” In Urbanization in the World-Economy, ed. M. Timberlake. New York: Academic Press.

Tobler, Waldo. 1969. "Satellite Confirmation of Settlement Size Coefficients." Area I: 30-34.

Todd, Graham. 1995. “Going Global in the semiperiphery: world cities as political        projects, the case of Toronto.” In World Cities in a World-System, ed. Paul L. Knox and Peter J. Taylor.  New York, New York: Cambridge University Press.

UEM  Urban Environmental Monitoring 2000 “Global Urban Monitoring with ASTER.”

Ward, D., S.R. Phinn, and A.T. Murray. 2000. "Monitoring Growth in Rapidly Urbanizing Areas Using Remotely Sensed Data." The Professional Geographer 52:371-385.

Weber, Christiane 2001 “Urban agglomeration delimitation using remote sensing

            data” Pp. 145-160 in Donnay et al.

Weeks, John R. 2002. Population: An Introduction to Concepts and Issues: 8th Edition. Belmont, CA: Wadsworth Publishing Co.

—. 2003a. "Remote Sensing." Pp. 837-839 in Encyclopedia of Population, vol. 2, edited by P. Demeny and G. McNicoll. New York: MacMillan Reference USA.

—. 2003b. "The Role of Spatial Analysis in Demographic Research." in Spatially Integrated Social Science: Examples in Best Practice, edited by M. F. Goodchild and D. G. Janelle. New York: Oxford University Press.

—. 2003c. "Using Remote Sensing and Geographic Information Systems to Identify the Underlying Properties of Urban Environments." Pp. Ch 17 in New Forms of Urbanization: Conceptualizing and Measuring Human Settlement in the Twenty-first Century, edited by A. G. Champion and G. Hugo. London: Ashgate Publishing Limited.

Weeks, John R., M. Saad Gadalla, Tarek Rashed, James Stanforth, and Allan G. Hill. 2000. "Spatial Variability in Fertility in Menoufia, Egypt, Assessed Through the Application of Remote Sensing and GIS Technologies." Environment and Planning A 32:695-714.

Weeks, John R., Arthur Getis, Allan G. Hill, M. Saad Gadalla, and Tarek Rashed. 2004 (forthcoming). "The Fertility Transition in Egypt: Intra-Urban Patterns in Cairo." Annals of the Association of American Geographers.

Weeks, John R., Dennis Larson, and Tarek Rashed. 2003. "Contrast or Continuum:  The Creation and Application of an Urban Gradient Index." Annual Meeting of the Population Association of America, New Orleans.

Wilkinson, David. 1992. “Cities, civilizations and oikumenes: I.” Comparative Civilizations Review 27: 51-87 (Fall).

Wu, Changsan and Alan T. Murray. 2003. "Estimating Impervious Surface Distribution by Spectral Mixture Analysis." Remote Sensing of Environment 84:493-505.

Yeoh, Brenda. 1999. “Global/Globalizing Cities.” Progress In Human Geography 23, 4: 607-616.

Yuan, Yew, Richard M. Smith, and W. Fredrick Limp. 1997. “Remodeling census population with spatial information from Landsat TM imagery.” Computers, Environment, and Urban Systems 21(3/4):245-258